* Only bind loopback addresses when binding to local
Today when binding to local (the default) we bind to any address that is
a loopback address, or any address on an interface that declares itself
as a loopback interface. Yet, not all addresses on loopback interfaces
are loopback addresses. This arises on macOS where there is a link-local
address assigned to the loopback interface (fe80::1%lo0) and in Docker
services where virtual IPs of the service are assigned to the loopback
interface (docker/libnetwork#1877). These situations cause problems:
- because we do not handle the scope ID of a link-local address, we end
up bound to an address for which publishing of that address does not
allow that address to be reached (since we drop the scope)
- the virtual IPs in the Docker situation are not loopback addresses,
they are not link-local addresses, so we end up bound to interfaces
that cause the bootstrap checks to be enforced even though the
instance is only bound to local
We address this by only binding to actual loopback addresses, and skip
binding to any address on a loopback interface that is not a loopback
address. This lets us simplify some code where in the bootstrap checks
we were skipping link-local addresses, and in writing the ports file
where we had to skip link-local addresses because again the formatting
of them does not allow them to be connected to by another node (to be
clear, they could be connected to via the scope-qualified address, but
that information is not written out).
Relates #28029
This commit fixes a flubbed assertion in the Wildfly build file; the
intention was to assert that we have only set the value of the
management port once except the assertion that the management port was
equal to zero when finding a matching log line was left off.
The Wildfly tests previously needed hardcoded ports which would
sometimes already be in use leading to Wildfly test failures because the
server could not bind during startup. We had to wait until some upstream
changes (wildfly/wildfly#9943) and (wildfly/wildfly-core#2390) were
released before we could use ephemeral ports in these tests. These
changes are released now so this commit changes the Wildfly tests to use
ephemeral ports when starting Wildfly.
Relates #28040
This commit disables the nio transport as an option for the test
transport in integration tests. This is because it does not currently
run properly in intellij due to socket permissions. It should be
reenabled once #27881 is merged (and the proper permissions are added).
With this commit we upgrade the Gradle Shadow plugin that is used in our
benchmarks to version 2.0.2. This version does not use APIs that are
deprecated in Gradle 4.x.
- Introduce index level settings to control the maximum number of terms
that can be used in a Terms Query
- Throw an error if a request exceeds this max number
Closes#18829
* Adds task dependenciesInfo to BuildPlugin to generate a CSV file with dependencies information (name,version,url,license)
* Adds `ConcatFilesTask.groovy` to concatenates multiple files into one
* Adds task `:distribution:generateDependenciesReport` to concatenate `dependencies.csv` files into a single file (`es-dependencies.csv` by default)
# Examples:
$ gradle dependenciesInfo :distribution:generateDependenciesReport
## Use `csv` system property to customize the output file path
$ gradle dependenciesInfo :distribution:generateDependenciesReport -Dcsv=/tmp/elasticsearch-dependencies.csv
## When branch is not master, use `build.branch` system property to generate correct licenses URLs
$ gradle dependenciesInfo :distribution:generateDependenciesReport -Dbuild.branch=6.x -Dcsv=/tmp/elasticsearch-dependencies.csv
This change fixes a bug when a keyword term in the `after` key is not present in the shard.
In this case the global ord of the document values are compared with the insertion point of the
`after` keyword and values that are equal to the insertion point should be considered "after" the top value.
We disable these tests because the JVM flags we use will not start with
JDK 10. Since we do not support running these old versions on newer JDKs
anyway, this commit disables testing of these when running on JDK 10.
In PR #27965, we set the global checkpoint from the translog in a store
recovery. However, we set after an engine is opened. This causes the
global checkpoint assertion in TranslogWriter violated as if we are
forced to close the engine before we set the global checkpoint. A
closing engine will close translog which in turn read the current global
checkpoint; however it is still unassigned and smaller than the initial
global checkpoint from translog.
Closes#27970
Today we always recover a primary from the last commit point. However
with a new deletion policy, we keep multiple commit points in the
existing store, thus we have chance to find a good starting commit
point. With a good starting commit point, we may be able to throw away
stale operations. This PR rollbacks a primary to a starting commit then
recovering from translog.
Relates #10708
Non-Peer recoveries should restore the global checkpoint rather than wait for the activation of the primary. This brings us a step closer to a universe where a recovered shard always has a valid global checkpoint. Concretely:
1) Recovery from store can read the checkpoint from the translog
2) Recovery from local shards and snapshots can set the global checkpoint to the local checkpoint as this is the only copy of the shard.
3) Recovery of an empty shard can set it to `NO_OPS_PERFORMED`
Peer recoveries will follow but require more work and thus will have their own PR.
I also used the moment to clean up `IndexShard`'s api around starting the engine and doing recovery from the translog. The current naming are a relic of the past and don't align with the current naming schemes in the engine.
This assertion does not hold if a shard is recovered from an empty store
but failed then retries. Moreover, if the openMode is CREATE_INDEX_*, we
pass CREATE mode to the IndexWriterConfig to create a new index and
overwrite the existing one.
Closes#27960
We disabled the Javadoc task on JDK 10 due to an apparent bug in Javadoc
generation on JDK 10. However, the client JAR task sets up its own
Javadoc task for client JARs (because it merely copies the non-client
Javadoc JAR). This commit diables that task too, since the Javadocs for
the non-client JAR will not exist.
Relates #27962
When a new snapshot is created it is added to the cluster state as a
snapshot-in-progress in INIT state, and the initialization is kicked
off in a new runnable task by SnapshotService.beginSnapshot(). The
initialization writes multiple files before updating the cluster state
to change the snapshot-in-progress to STARTED state. This leaves a
short window in which the snapshot could be deleted (let's say, because
the snapshot is stuck in INIT or because it takes too much time to
upload all the initialization files for all snapshotted indices). If
the INIT snapshot is deleted, the snapshot-in-progress becomes ABORTED
but once the initialization in SnapshotService.beginSnapshot() finished
it is change back to STARTED state again.
This commit avoids an ABORTED snapshot to be started if it has been
deleted during initialization. It also adds a test that would have failed
with the previous behavior, and changes few method names here and there.
This commit changes some Azure tests so that they do not rely on
MockZenPing and TestZenDiscovery anymore, but instead use a mocked
AzureComputeService that exposes internal test cluster nodes as if
they were real Azure nodes.
Related to #27859Closes#27917, #11533
There appears to be a bug in JDK 10 for generating Javadocs with some
nested anonymous classes. This commit disables these on JDK 10 until the
upstream issue is resolved.
Relates #27952
Today we don't persist the global checkpoint when finishing a peer
recovery even though we advance an in memory value. This commit persists
the global checkpoint in RecoveryTarget#finalizeRecovery.
Closes#27861
Search response listeners should not be exposed in search plugin.
Support was added but reverted right after (not present in any release).
Though the SearchPlugin still contains a default definition for search response listeners
due to a broken revert. This change removes this extension point that is basically no-op.
In the case of nested aggregator this allows it to push down buffered child docs down to children aggregator.
Before this was done as part of the `NestedAggregator#getLeafCollector(...)`, but by then the children aggregators
have already moved on to the next segment and this causes incorrect results to be produced.
Closes#27912
* Limit the analyzed text for highlighting
- Introduce index level settings to control the max number of character
to be analyzed for highlighting
- Throw an error if analysis is required on a larger text
Closes#27517
* Splits nio project into two for eclipse build only
https://github.com/elastic/elasticsearch/pull/27801 introduced a new gradle project `:libs:elasticsearch-nio` which creates cyclical project dependencies when importingthe projects into Eclipse.
This change applies the same trick as we have for the core project where, and building for Eclipse, splits the `:libs:elasticsearch-nio` project into `:libs:elasticsearch-nio` which points to `src/main` and `:libs:elasticsearch-nio-tests` which points to `src/test`. This prevents cyclical project dependencies in Eclipse arising from the fact that eclipse does not separate compile/runtime dependencies from test dependencies.
* Removes integTest bits since there are none
Lucene TopDocs collector are now able to early terminate the collection
based on the index sort. This change plugs this new functionality directly in the
query phase instead of relying on a dedicated early terminating collector.
These templates were generated with 5.0. We need those generated with 6.0 since
we do not guarantee compatibility with previous versions of the template anyway.
I removed the winlogbeat template which is a bit harder to generate as it requires
Windows, since we do not aim to be exhaustive.
This is related to #27802. This commit adds a jar called
elasticsearch-nio that contains the base nio classes that will be used
for the tcp nio transport and eventually the http nio transport.
The jar does not depend on elasticsearch:core, so all references to core
have been removed.
JDK 10 has gone fully-modular. This means that:
- when targeting JDK 8 with a JDK 10 compiler, we have to use the full
profile
- when targeting JDK 10 with a JDK 10 compiler, we have to use
-add-modules java.base
Today we only target JDK 8 (our minimum compatibility version) so we
need to change the compiler flags conditional on using a JDK 10
compiler. This commit does that.
Relates #27884
Tanguy Leroux
Jason Tedor
Gytis Šk
Timoshenko
Tim Brooks
Yannick Welsch
Mayya Sharipova
Daniel Mitterdorfer
Shaunak Kashyap
Vlad Holubiev
Doug Turnbull
Maxime Gréau
Jim Ferenczi
Nhat Nguyen
Boaz Leskes
Ryan Ernst
Alexander Reelsen
olcbean
Martijn van Groningen
Colin Goodheart-Smithe
Jim Ferenczi
Adrien Grand
Andy Bristol
Sian Lerk Lau